Merge pull request #151 from mempool/mononaut/asset-search

src/elements/mod.rs

@@ -9,7 +9,7 @@ mod registry;
 
 use asset::get_issuance_entropy;
 pub use asset::{lookup_asset, LiquidAsset};
-pub use registry::{AssetRegistry, AssetSorting};
+pub use registry::{AssetMeta, AssetRegistry, AssetSorting};
 
 #[derive(Serialize, Deserialize, Clone)]
 pub struct IssuanceValue {

src/elements/registry.rs

@@ -14,6 +14,8 @@ use crate::errors::*;
 // (in number of hex characters, not bytes)
 
 const DIR_PARTITION_LEN: usize = 2;
+const SEARCH_SORT_CANDIDATE_LIMIT: usize = 2000;
+
 pub struct AssetRegistry {
     directory: path::PathBuf,
     assets_cache: HashMap<AssetId, (SystemTime, AssetMeta)>,
@@ -53,6 +55,39 @@ impl AssetRegistry {
         )
     }
 
+    pub fn search(&self, query: &str, limit: usize) -> Vec<AssetEntry<'_>> {
+        let query = query.trim();
+        if query.is_empty() || limit == 0 {
+            return vec![];
+        }
+
+        let (mut results, candidates) = search_by(
+            self.assets_cache
+                .iter()
+                .map(|(asset_id, (_, metadata))| (asset_id, metadata)),
+            query,
+            limit,
+            |metadata| metadata.ticker.as_deref(),
+        );
+
+        if results.len() < limit {
+            let (name_matches, candidates) =
+                search_by(candidates, query, limit - results.len(), |metadata| {
+                    Some(&metadata.name)
+                });
+            results.extend(name_matches);
+
+            if results.len() < limit {
+                let (domain_matches, _) =
+                    search_by(candidates, query, limit - results.len(), AssetMeta::domain);
+                results.extend(domain_matches);
+            }
+        }
+
+        results.truncate(limit);
+        results
+    }
+
     pub fn fs_sync(&mut self) -> Result<()> {
         for entry in fs::read_dir(&self.directory).chain_err(|| "failed reading asset dir")? {
             let entry = entry.chain_err(|| "invalid fh")?;
@@ -126,7 +161,7 @@ pub struct AssetMeta {
 }
 
 impl AssetMeta {
-    fn domain(&self) -> Option<&str> {
+    pub(crate) fn domain(&self) -> Option<&str> {
         self.entity["domain"].as_str()
     }
 }
@@ -192,3 +227,72 @@ fn lc_cmp_opt(a: &Option<String>, b: &Option<String>) -> cmp::Ordering {
         .map(|a| a.to_lowercase())
         .cmp(&b.as_ref().map(|b| b.to_lowercase()))
 }
+
+fn search_by<'a, I, F>(
+    candidates: I,
+    query: &str,
+    limit: usize,
+    field: F,
+) -> (Vec<AssetEntry<'a>>, Vec<AssetEntry<'a>>)
+where
+    I: IntoIterator<Item = AssetEntry<'a>>,
+    F: Fn(&AssetMeta) -> Option<&str>,
+{
+    let mut matches = vec![];
+    let mut remaining = vec![];
+
+    for (asset_id, metadata) in candidates {
+        let position = field(metadata).and_then(|field| {
+            // registry fields are ascii, so we don't need full unicode case-folding
+            ascii_ci_find(field, query).map(|position| (position, field))
+        });
+
+        if let Some((position, field)) = position {
+            if matches.len() >= SEARCH_SORT_CANDIDATE_LIMIT {
+                continue;
+            }
+            matches.push((position, field, asset_id, metadata));
+        } else {
+            remaining.push((asset_id, metadata));
+        }
+    }
+
+    matches.sort_unstable_by(|a, b| {
+        a.0.cmp(&b.0)
+            .then_with(|| ascii_ci_cmp(a.1, b.1))
+            .then_with(|| a.2.cmp(b.2))
+    });
+
+    (
+        matches
+            .into_iter()
+            .take(limit)
+            .map(|(_, _, asset_id, metadata)| (asset_id, metadata))
+            .collect(),
+        remaining,
+    )
+}
+
+// zero-allocation case-insensitive ASCII substring search
+// returns the byte offset of the first match
+fn ascii_ci_find(haystack: &str, needle: &str) -> Option<usize> {
+    let (haystack, needle) = (haystack.as_bytes(), needle.as_bytes());
+    if needle.is_empty() {
+        return Some(0);
+    }
+    haystack
+        .windows(needle.len())
+        .position(|window| window.eq_ignore_ascii_case(needle))
+}
+
+// zero-allocation case-insensitive ASCII string comparison
+fn ascii_ci_cmp(a: &str, b: &str) -> cmp::Ordering {
+    let (a, b) = (a.as_bytes(), b.as_bytes());
+    for i in 0..a.len().min(b.len()) {
+        match a[i].to_ascii_lowercase().cmp(&b[i].to_ascii_lowercase()) {
+            cmp::Ordering::Equal => continue,
+            ord => return ord,
+        }
+    }
+    a.len().cmp(&b.len())
+}

src/new_index/query.rs

@@ -14,7 +14,7 @@ use crate::util::{is_spendable, BlockId, Bytes, TransactionStatus};
 #[cfg(feature = "liquid")]
 use crate::{
     chain::{asset::AssetRegistryLock, AssetId},
-    elements::{lookup_asset, AssetRegistry, AssetSorting, LiquidAsset},
+    elements::{lookup_asset, AssetMeta, AssetRegistry, AssetSorting, LiquidAsset},
 };
 
 const FEE_ESTIMATES_TTL: u64 = 60; // seconds
@@ -271,6 +271,15 @@ impl Query {
         )
     }
 
+    #[cfg(feature = "liquid")]
+    pub fn lookup_registry_asset(&self, asset_id: &AssetId) -> Result<Option<AssetMeta>> {
+        let asset_db = self
+            .asset_db
+            .as_ref()
+            .chain_err(|| "asset registry unavailable")?;
+        Ok(asset_db.read().unwrap().get(asset_id).cloned())
+    }
+
     #[cfg(feature = "liquid")]
     pub fn list_registry_assets(
         &self,
@@ -298,4 +307,27 @@ impl Query {
             .collect::<Result<Vec<_>>>()?;
         Ok((total_num, results))
     }
+
+    #[cfg(feature = "liquid")]
+    pub fn search_registry_assets<T, F>(
+        &self,
+        search: &str,
+        limit: usize,
+        mut map: F,
+    ) -> Result<Vec<T>>
+    where
+        F: FnMut(&AssetId, &AssetMeta) -> T,
+    {
+        let asset_db = self
+            .asset_db
+            .as_ref()
+            .chain_err(|| "asset registry unavailable")?;
+        Ok(asset_db
+            .read()
+            .unwrap()
+            .search(search, limit)
+            .into_iter()
+            .map(|(asset_id, metadata)| map(asset_id, metadata))
+            .collect())
+    }
 }

src/rest.rs

@@ -35,7 +35,7 @@ use hyperlocal::UnixServerExt;
 use std::{cmp, fs};
 #[cfg(feature = "liquid")]
 use {
-    crate::elements::{peg::PegoutValue, AssetSorting, IssuanceValue},
+    crate::elements::{peg::PegoutValue, AssetMeta, AssetSorting, IssuanceValue},
     elements::{
         confidential::{Asset, Nonce, Value},
         encode, AssetId,
@@ -59,6 +59,12 @@ const MULTI_ADDRESS_LIMIT: usize = 300;
 const ASSETS_PER_PAGE: usize = 25;
 #[cfg(feature = "liquid")]
 const ASSETS_MAX_PER_PAGE: usize = 100;
+#[cfg(feature = "liquid")]
+const ASSETS_SEARCH_DEFAULT_LIMIT: usize = 15;
+#[cfg(feature = "liquid")]
+const ASSETS_SEARCH_MAX_LIMIT: usize = 100;
+#[cfg(feature = "liquid")]
+const ASSETS_SEARCH_MAX_QUERY_LEN: usize = 64;
 
 const TTL_LONG: u32 = 157_784_630; // ttl for static resources (5 years)
 const TTL_SHORT: u32 = 10; // ttl for volatie resources
@@ -132,6 +138,32 @@ impl BlockValue {
     }
 }
 
+#[cfg(feature = "liquid")]
+#[derive(Serialize)]
+struct AssetRegistrySearchResult {
+    asset_id: AssetId,
+    name: String,
+
+    #[serde(skip_serializing_if = "Option::is_none")]
+    ticker: Option<String>,
+
+    #[serde(skip_serializing_if = "Option::is_none")]
+    domain: Option<String>,
+}
+
+#[cfg(feature = "liquid")]
+impl AssetRegistrySearchResult {
+    fn new(asset_id: &AssetId, meta: &AssetMeta) -> Self {
+        let domain = meta.domain().map(String::from);
+        Self {
+            asset_id: *asset_id,
+            name: meta.name.clone(),
+            ticker: meta.ticker.clone(),
+            domain,
+        }
+    }
+}
+
 /// Calculate the difficulty of a BlockHeader
 /// using Bitcoin Core code ported to Rust.
 ///
@@ -1774,6 +1806,38 @@ fn handle_request(
             json_response(recent, TTL_MEMPOOL_RECENT)
         }
 
+        #[cfg(feature = "liquid")]
+        (&Method::GET, Some(&"assets"), Some(&"registry"), Some(&"search"), None, None) => {
+            let search = query_params.get("q").map(|q| q.trim()).unwrap_or("");
+            let assets = if search.is_empty() {
+                vec![]
+            } else if search.chars().count() > ASSETS_SEARCH_MAX_QUERY_LEN {
+                return Err(HttpError(
+                    StatusCode::BAD_REQUEST,
+                    "search query too long".to_string(),
+                ));
+            } else {
+                let limit = query_params
+                    .get("limit")
+                    .and_then(|n| n.parse::<usize>().ok())
+                    .unwrap_or(ASSETS_SEARCH_DEFAULT_LIMIT)
+                    .min(ASSETS_SEARCH_MAX_LIMIT);
+
+                query
+                    .search_registry_assets(search, limit, AssetRegistrySearchResult::new)
+                    .map_err(|e| {
+                        HttpError(StatusCode::SERVICE_UNAVAILABLE, e.description().to_string())
+                    })?
+            };
+
+            Ok(Response::builder()
+                // Disable caching because we don't currently support caching with query string params
+                .header("Cache-Control", "no-store")
+                .header("Content-Type", "application/json")
+                .body(Body::from(serde_json::to_string(&assets)?))
+                .unwrap())
+        }
+
         #[cfg(feature = "liquid")]
         (&Method::GET, Some(&"assets"), Some(&"registry"), None, None, None) => {
             let start_index: usize = query_params
@@ -1800,6 +1864,21 @@ fn handle_request(
                 .unwrap())
         }
 
+        #[cfg(feature = "liquid")]
+        (&Method::GET, Some(&"assets"), Some(&"registry"), Some(asset_str), None, None) => {
+            let asset_id = AssetId::from_str(asset_str)?;
+            let registry_entry = query
+                .lookup_registry_asset(&asset_id)
+                .map_err(|e| {
+                    HttpError(StatusCode::SERVICE_UNAVAILABLE, e.description().to_string())
+                })?
+                .ok_or_else(|| {
+                    HttpError::not_found("Asset id not found in registry".to_string())
+                })?;
+
+            json_response(registry_entry, TTL_SHORT)
+        }
+
         #[cfg(feature = "liquid")]
         (&Method::GET, Some(&"asset"), Some(asset_str), None, None, None) => {
             let asset_id = AssetId::from_str(asset_str)?;